AI in Investigative Journalism: Joris Veerbeek's Pioneering Work at De Groene Amsterdammer
utrecht, woensdag, 3 september 2025.
Joris Veerbeek, a researcher at Utrecht University, uses artificial intelligence to analyse hundreds of thousands of tweets for hate directed at female politicians. His work at De Groene Amsterdammer demonstrates how AI can enrich and make journalism more efficient without replacing human interpretation. Veerbeek emphasises that journalists do not need to become programmers, but they should understand how AI models work.
A Pioneer Role in Journalism
Joris Veerbeek, a researcher at Utrecht University, combines his academic expertise with practical experience at De Groene Amsterdammer. He is pursuing his doctorate on the use of generative AI in investigative journalism and applies this technology by analysing hundreds of thousands of tweets for hate directed at female politicians [1]. Veerbeek works three days a week at the university and two days at De Groene Amsterdammer, where he uses AI to sift through large datasets and find relevant information more quickly [1].
Methodology and Applications
For his research on hate tweets directed at female politicians, Veerbeek trained an AI model to analyse 300,000 tweets [1]. This process involved a team of four people who manually labelled 10,000 tweets to train the model [1]. Training the AI model took three months and required precise collaboration and oversight to ensure the reliability of the results [1]. The flexibility of new AI models, such as those from OpenAI, requires less data and time for specific tasks, allowing journalists to work faster and more efficiently [1].
Impact on News Production and Consumption
The use of AI in journalism has both advantages and potential drawbacks. One of the main advantages is the speed at which large amounts of data can be analysed. This enables journalists to quickly gain insights into complex trends and patterns, such as identifying hate speech on social media [1]. Additionally, AI helps automate routine tasks, giving journalists more time to focus on in-depth investigations and stories [1].
Ethical Considerations and Human Interpretation
Despite the advantages, there are also ethical considerations when using AI in journalism. Veerbeek stresses that journalists do not need to become programmers, but they should understand how AI models work and how they can be deployed [1]. Generative AI is not intended for indiscriminate use in every application, and human interpretation remains essential for filtering and interpreting data [1]. Data labelling is approached deliberately and calmly, with regular breaks to ensure the well-being of the team [1].
Future Perspectives
Veerbeek is optimistic about the future of AI in journalism. He believes that the technology will evolve and offer even more possibilities for in-depth investigations and stories [1]. However, the human factor remains crucial. Journalists must not only be technically skilled but also think critically and act ethically [1]. The first scientific paper on the use of AI in journalism was published earlier this year, demonstrating that the field is rapidly developing and growing [1].